基于临床特征的胆道恶性肿瘤术前鉴别模型构建及表观性能评价OA

Objective To develop a preoperative model for differentiating biliary tract cancers(BTC)from benign biliary diseases in patients undergoing biliary surgery and to evaluate its apparent performance.Methods This retrospective study included 90 patients with BTC admitted to the hospital from January 2024 to January 2026 as the case group and 90 patients with benign biliary diseases without malignancy who underwent biliary surgery during the same period as the control group.Clinical characteristics,biochemical indicators,and tumor markers[carbohydrate antigen 19-9(CA19-9),carbohydrate antigen 125(CA125),and carcinoembryonic antigen(CEA)]were compared between the two groups.Variables showing statistically significant differences in univariate analysis were entered into a multivariable logistic regression model to identify independent discriminative factors associated with BTC case status.A nomogram for preoperative differential diagnosis was constructed based on the multivariable regression results,and receiver operating characteristic(ROC)curve analysis was performed to assess the model's discriminative performance.Results Patients in the case group were older and had higher proportions of cirrhosis,hypertension,hepatitis B,diabetes mellitus,bile duct stones,HDL-C≤0.83 mmol/L,elevated CA19-9,elevated CA125,and elevated CEA than those in the control group(P＜0.05).After adjustment for confounding factors,multivariable logistic regression analysis showed that age,cirrhosis,hypertension,hepatitis B,diabetes mellitus,bile duct stones,HDL-C≤0.83 mmol/L,elevated CA19-9,elevated CA125,and elevated CEA were independent discriminative factors for distinguishing BTC from benign biliary diseases(P＜0.05).A preoperative diagnostic nomogram was constructed using the"rms"package in R based on these variables.ROC curve analysis showed that the model had an apparent AUC of 0.915(95%CI:0.876～0.955),with a specificity of 0.833,sensitivity of 0.878,and Youden index of 0.711.In this dataset,the Hosmer-Lemeshow test suggested acceptable internal model fit(P＞0.05),but this does not indicate reliable calibration in the general population or in external populations.The calibration curve in the development sample showed agreement between model-predicted probabilities and the observed case distribution within this cohort.Conclusion Age,cirrhosis,hypertension,hepatitis B,diabetes mellitus,bile duct stones,low HDL-C,CA19-9,CA125,and CEA are independent discriminative factors for distinguishing BTC from benign biliary lesions among patients undergoing biliary surgery.The nomogram-based preoperative discrimination model constructed from these variables showed good apparent discrimination within the present cohort and may assist clinicians in identifying patients with suspected BTC among surgical candidates with biliary disease.